CEDAR

Inspiration

Carotid artery stenosis is the leading cause of ischemic stroke in America, and there is a 20% higher risk of stroke-related deaths in rural America. Current at-home ultrasound devices also introduce significant levels of variability depending on how patients orient the probe. The question is, how can high-risk patients, living in rural areas, have more frequent and convenient ultrasound check-ups with their physicians?

What it does

CEDAR is a flexible wearable device that contains two newly developed ultrasound patches, collecting full images of each carotid artery. These patches contain newly developed piezoelectric material to provide clear, accurate images. Rather than relying on patients inconsistently taking their own ultrasounds, CEDAR is a patch that rests on the neck, takes the ultrasound, and sends this data to patients’ personal devices via Bluetooth. In addition to helping patients, CEDAR will help physicians through an interface with automatic detection of the artery of interest and updating the patient’s EHR. Summary metrics will act as a pre-screening method for physicians. CEDAR is augmented by a deep-learning model for image processing. Pre-processed images will be used to train the FCN-32 model, which accepts the unprocessed images from the continuous image recording and predicts the carotid artery volume, yielding waveforms of key cardiac performance indices including stenosis severity. This technology enables dynamic wearable monitoring of cardiac performance with substantially improved accuracy in various environments. Overall, we aim to alleviate the financial and emotional stress of stroke patients, who currently spend up to $100,000 per year on monitoring and treatment.

How I built it

CEDAR leverages a newly developed stretchable ultrasonic array capable of serial, non-invasive, three-dimensional imaging of tissues as deep as four centimeters below the surface of human skin, at a spatial resolution of 0.5 millimeters. There is a flexible patch made of silicone rubber, embedded with five ultrasound arrays made from a new piezoelectric material that the researchers developed. The arrays are positioned in the shape of a cross, which allows the patch to image the entire bladder, which is about 12 by 8 centimeters (Hu et al., 2023).

Our model is trained on pre-existing processed ultrasound images to optimize spatial resolutions, location accuracies, and signal-to-noise ratios. Pre-processed images will be used to train the FCN-32 model, which accepts the unprocessed images from the continuous image recording and predicts the carotid artery volume, yielding waveforms of key artery performance indices including stenosis severity.

Challenges I ran into

• Adapting the initial cardiac application of the ultrasound technology to carotid artery stenosis, which tracks different waveforms of performance

What I learned

• Deep learning image processing
• Ultrasonic technology

What's next for CEDAR

• Development of wearable patch device
• Integration into physician EHR systems
• Create a user interface with features such as

Built With

• arduino
• fcn-32